A human chorionic gonadotropin content detection system

Abstract

The present invention discloses a human chorionic gonadotropin (HCG) content detection system, comprising: an extreme feedback module internally provided with a color development acceleration device for accelerating the color development of a test strip until the color development area displays the extreme color; a first acquisition module for collecting test strip images of each test strip displaying the extreme color; a difference calculation module for acquiring the RGB data of each pixel in the first color development area and establishing a difference function; a second acquisition module for collecting real-time images of each test strip; a model training module for using the labeled real-time image as input and the first color development area as output, retraining the initial model to obtain a feature labeling model; a content output module for inputting the unlabeled real-time image into the feature labeling model to obtain a feature display area, adjusting the content detection model according to the difference function to obtain an optimized detection model, and inputting the RGB data of each pixel in the feature display area into the optimized detection model to obtain the hormone content. The present invention improves the accuracy of human chorionic gonadotropin content detection.

Description

Technical Field

[0001] The present invention relates to the technical field of immunochromatographic concentration detection, in particular to a human chorionic gonadotropin content detection system. Background Art

[0002] Human chorionic gonadotropin (HCG) is a glycoprotein secreted by trophoblast cells of the placenta. It is composed of α- and β-dimers. Currently, the immunochromatographic method is commonly used to measure hCG levels. After the immunochromatographic test, the traditional method is to manually compare the immunochromatographic test strip with a standard color chart to determine the concentration. However, manual comparison is subject to significant subjective factors and is prone to misjudgment. Therefore, machine-based comparison is currently more commonly used. This involves placing the immunochromatographic test strip in a camera to capture an RGB image. Analysis software is then used to import this RGB image and analyze the RGB data of a specific area. The RGB data is then determined to determine the corresponding concentration based on the color interval. This machine-based comparison method significantly improves detection accuracy compared to manual comparison, effectively overcoming its shortcomings. However, current machine-based comparison methods generally use various OpenCV-based edge detection algorithms (including Sobel, Canny, and Laplacian) to detect grayscale images. However, during the shooting and imaging process, the existing technology ignores the color value deviation of the shooting lens itself in the RGB image shooting and imaging process, resulting in inevitable systematic errors in the final human chorionic gonadotropin content detection results, resulting in insufficient accuracy in the detection of human chorionic gonadotropin content. Summary of the Invention

[0003] In view of the deficiencies in the prior art, the object of the present invention is to provide a human chorionic gonadotropin content detection system for improving the accuracy of human chorionic gonadotropin content detection.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a human chorionic gonadotropin content detection system, comprising:

[0005] The extreme feedback module is internally provided with a color development acceleration device and a plurality of placement slots. The placement slots are used to place the human chorionic gonadotropin test strips after immunochromatography. The color development acceleration device is used to accelerate the color development of each test strip until the color development area displays the extreme color known by the RGB data;

[0006] a first acquisition module, connected to the extreme feedback module, for acquiring a test strip image of each of the test strips displaying an extreme color, and manually marking a first color-developed area;

[0007] a difference calculation module, connected to the first acquisition module, configured to obtain the image RGB data of each pixel in the first color rendering area, and establish a difference function based on the difference between the known RGB data of the limit color and the image RGB data of each pixel;

[0008] a second acquisition module, connected to the extreme feedback module, for acquiring real-time images of each test strip, and selecting a portion of the real-time image to manually mark the first color-developing area;

[0009] a model training module, connected to the second acquisition module, configured to introduce an initial model, take the marked real-time image as input, take the marked first color region as output, and retrain the initial model to obtain a feature labeling model;

[0010] The content output module is connected to the model training module and the difference calculation module, and is used to input the unlabeled real-time image into the feature labeling model to obtain a feature display area, and adjust the weight parameters of the pre-trained content detection model according to the difference function to obtain an optimized detection model, and then input the RGB data of each pixel point in the feature display area into the optimized detection model to obtain the hormone content.

[0011] Furthermore, the extreme feedback module includes a test strip placement box, an accelerated reaction chamber is provided in the test strip placement box, each placement slot is opened on the bottom inner wall of the accelerated reaction chamber, the color development acceleration device includes a disc-shaped shell, a ventilation shell and a lifting mechanism, the upper end of the lifting mechanism is fixed to the top inner wall of the accelerated reaction chamber, the driving output end of the lifting mechanism is fixedly connected to the ventilation shell, the lifting mechanism is used to drive the ventilation shell to lift vertically, an air inlet channel is provided inside the ventilation shell, and an air inlet channel is provided inside the air inlet channel. The fan, the disc-shaped shell is fixedly connected to the lower end of the ventilation shell, and a plurality of air outlet channels are provided inside the disc-shaped shell. The air inlets of each air outlet channel are connected to the air inlet channel, and a heating resistance wire is provided inside each air outlet channel. The air outlet of each air outlet channel faces the sample placement area of ​​each test strip, and the air outlet of each air outlet channel is provided with a temperature detection device, a wind speed detection device and a distance detection device, which are respectively used to detect the temperature and wind speed at the air outlet and the acceleration distance between the air outlet and the test strip in real time.

[0012] Furthermore, the interior of the accelerated reaction chamber is hemispherical, the inner wall of the side end of the accelerated reaction chamber is provided with a thermal insulation coating, and a plurality of vertically spaced gas flow channels are evenly opened on the thermal insulation coating.

[0013] Furthermore, a timing unit and a control unit are provided in the test strip placement box, and the control unit is respectively connected to the lifting mechanism, each heating resistance wire, the air intake fan, each temperature detection device, each wind speed detection device and the timing unit;

[0014] The timing unit is used to start timing after the test strip is placed in the placement slot, and stop timing to generate a limit time after the test strip displays the limit color. The control unit establishes an objective function based on the temperature, wind speed, acceleration distance and corresponding limit time at each air outlet, solves the minimum value of the objective function, and outputs the corresponding temperature, wind speed, and acceleration distance as the optimal combination when the objective function is minimum and saves it.

[0015] Furthermore, the objective function is configured as:

[0016]

[0017]

[0018] Wherein, f(x) is used to represent the objective function;

[0019] T i Used to indicate the limit time;

[0020] T0 is used to represent the preset standard time, which is a positive number;

[0021] k1, k2, and k3 are used to represent a preset first coefficient, a second coefficient, and a third coefficient, respectively, where the first coefficient, the second coefficient, and the third coefficient are all constants;

[0022] T P Used to indicate the temperature at the air outlet;

[0023] W i Used to indicate the wind speed at the air outlet;

[0024] I n Used to represent the acceleration spacing.

[0025] Furthermore, the difference function is configured as:

[0026]

[0027] Among them, D V Used to represent the difference function;

[0028] R maxi Known RGB data for representing the i-th pixel of the test strip in the first color development area that presents the extreme color;

[0029] R i Used to represent actual RGB data of the i-th pixel in the first color development area of ​​the test strip presenting the extreme color.

[0030] Furthermore, the initial model is a U-Net prediction model. In the process of retraining the U-Net prediction model based on the real-time image and the first color display area, a residual block and an adversarial network are superimposed on the U-Net prediction model. The residual block is used to improve the defect segmentation effect, and the adversarial network is used to supervise the output of the U-Net prediction model and perform feedback adjustment on the U-Net prediction model.

[0031] Furthermore, the content output module includes:

[0032] a superposition unit, configured to superimpose the difference function and the content detection model to obtain a superposition function;

[0033] A calculation unit, configured to find the minimum value of the difference function;

[0034] A correction unit is connected to the calculation unit and the superposition unit respectively, and is used to adjust the weight parameters of the content detection model according to the minimum value of the difference function when the minimum value of the difference function is obtained, and output the adjusted superposition function as the optimized detection model.

[0035] Furthermore, the first acquisition module and the second acquisition module both include preprocessing units, respectively used to eliminate invalid data, non-steady-state data and abnormal data in the test strip image to obtain the preprocessed test strip data.

[0036] Beneficial effects of the present invention:

[0037] The present invention sets an extreme feedback module and utilizes the extreme feedback module to accelerate the immunochromatographic reaction speed of human chorionic gonadotropin, thereby ensuring that the test strip can display extreme colors and improving the speed and stability of the test strip color development; furthermore, by collecting the image RGB data of each pixel point on the first color development area of ​​the collected test strip image, a difference function is established with the known RGB data of the extreme color and the difference therewith, and the difference function is used to adjust the weight parameter of the content detection function, thereby avoiding the influence of the color value deviation generated in the process of collecting the test strip image, thereby improving the accuracy of human chorionic gonadotropin content detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 Schematic diagram of the structure of the human chorionic gonadotropin content detection system of the present invention;

[0039] Figure 2It is an internal cross-sectional view of the test strip placement box of the present invention.

[0040] Figure numerals: 1. extreme feedback module; 11. timing unit; 12. control unit; 2. first acquisition module; 3. difference calculation module; 4. second acquisition module; 5. model training module; 6. content output module; 61. superposition unit; 62. calculation unit; 63. correction unit; 7. paper strip placement box; 71. acceleration reaction chamber; 72. placement slot; 73. disc-shaped shell; 74. ventilation shell; 75. air intake fan; 76. heating resistor wire; 77. temperature detection device; 78. wind speed detection device. DETAILED DESCRIPTION

[0041] The present invention will be described in further detail below with reference to the accompanying drawings and embodiments. Identical components are denoted by the same reference numerals. It should be noted that the terms "front," "rear," "left," "right," "upper," and "lower" used in the following description refer to directions in the accompanying drawings, and the terms "bottom," "top," "inner," and "outer" refer to directions toward or away from the geometric center of a particular component, respectively.

[0042] like Figure 1 and Figure 2 As shown, a human chorionic gonadotropin content detection system of this embodiment includes:

[0043] The extreme feedback module 1 is internally provided with a color development acceleration device and a plurality of placement slots 72. The placement slots 72 are used to place the human chorionic gonadotropin test strips after immunochromatography. The color development acceleration device is used to accelerate the color development of each test strip until the color development area displays the extreme color known by the RGB data.

[0044] The first acquisition module 2 is connected to the extreme feedback module 1 and is used to acquire the test strip image of each test strip showing the extreme color and manually mark the first color display area;

[0045] The difference calculation module 3 is connected to the first acquisition module 2 and is used to obtain the image RGB data of each pixel in the first color rendering area and establish a difference function based on the difference between the known RGB data of the extreme color and the image RGB data of each pixel;

[0046] The second acquisition module 4 is connected to the extreme feedback module 1 and is used to collect real-time images of each test strip and select part of the real-time image to manually mark the first color development area;

[0047] The model training module 5 is connected to the second acquisition module 4 and is used to introduce an initial model, and use the marked real-time image as input and the marked first color area as output, and retrain the initial model to obtain a feature labeling model;

[0048] The content output module 6 is connected to the model training module 5 and the difference calculation module 3, and is used to input the unlabeled real-time image into the feature labeling model to obtain the feature display area, and adjust the weight parameters of the pre-trained content detection model according to the difference function to obtain the optimized detection model, and then input the RGB data of each pixel point in the feature display area into the optimized detection model to obtain the hormone content.

[0049] Specifically, in this embodiment, both the first acquisition module 2 and the second acquisition module 4 can be ultra-high-definition cameras. Before performing immunochromatographic testing on human chorionic gonadotropin test strips, a difference function must first be obtained. The difference function acquisition process includes placing the immunochromatographic test strips in the extreme feedback module 1 and accelerating the color development of the color-developing area on the test strips using a color acceleration device until the test strips display the extreme color. The test strips displaying the extreme color are removed, and the first acquisition module 2 captures an image of the test strips. The difference calculation module 3 uses an OpenCV algorithm to obtain the RGB data of each pixel in the first color-developing area and subtracts it from the known RGB data of the extreme color to establish a difference function. After obtaining the difference function, the second acquisition module 4 captures real-time images of each test strip in real time. The captured real-time images are divided into two parts: one unlabeled and the other with the first color-developing area manually labeled as a training set. The model training module 5 inputs the labeled real-time images into the initial model and uses the first color-developing area as output to train a feature labeling model. The unlabeled real-time image is input into the feature labeling model, and the feature display area is finally output. The second display area is obtained by manually marking the real-time image corresponding to the feature display area again, and the weight parameter of the feature labeling model is adjusted by calculating the area difference between the second display area and the feature display area to improve the accuracy of the feature labeling model. The content detection model is initially a decision tree classifier model. By inputting the RGB data of each pixel point in the feature display area into the decision tree classifier model, the concentration content interval of human chorionic gonadotropin can be identified and output as the hormone content. In this embodiment, there are 10 concentration content intervals, namely 0-10ng / mL interval, 10-20ng / mL interval, 20-30ng / mL interval, 30-40ng / mL interval, 40-50ng / mL interval, 50-60ng / mL interval, 60-70ng / mL interval, 70-80ng / mL interval, 80-90ng / mL interval,

[0050] In the range of 90-100 ng / mL, this technical solution uses the difference function to adjust the weight parameters of the content detection model, avoiding the influence of color value deviation generated during the acquisition of test strip images, so that the hormone content output by the final optimized detection model is not affected by the systematic error during the acquisition process, thereby improving the accuracy of human chorionic gonadotropin content detection.

[0051] Preferably, the extreme feedback module 1 includes a test strip placement box 7, which is provided with an accelerated reaction chamber 71, and each placement slot 72 is opened on the bottom inner wall of the accelerated reaction chamber 71. The color development acceleration device includes a disc-shaped shell 73, a ventilation shell 74 and a lifting mechanism. The upper end of the lifting mechanism is fixed to the top inner wall of the accelerated reaction chamber 71, and the driving output end of the lifting mechanism is fixedly connected to the ventilation shell 74. The lifting mechanism is used to drive the ventilation shell 74 to lift vertically. The ventilation shell 74 is provided with an air inlet channel. There is an air inlet fan 75, and a disc-shaped shell 73 is fixedly connected to the lower end of the ventilation shell 74. A number of air outlet channels are provided inside the disc-shaped shell 73. The air inlets of each air outlet channel are connected to the air inlet channel. A heating resistor wire 76 is provided inside each air outlet channel. The air outlet of each air outlet channel faces the sample placement area of ​​each test strip. The air outlet of each air outlet channel is provided with a temperature detection device 77, a wind speed detection device 78 and a distance detection device, which are respectively used to detect the temperature and wind speed at the air outlet and the acceleration distance between the air outlet and the test strip in real time.

[0052] Specifically, in this embodiment, the air inlet fan 75 is used to blow air to each test strip through the air outlet to accelerate the penetration of the sample detection liquid and accelerate the color development. The heating resistor 76 is used to heat the air outlet channel so that the air blown out from the air outlet has a suitable temperature to accelerate the color development. In this embodiment, the optimal temperature at the air outlet is 37°, and the temperature at the air outlet cannot exceed 40°, otherwise it will seriously affect the results of the immunochromatographic reaction and even cause the inability to develop color. The lifting mechanism is used to adjust the distance between the vent and the test strip to achieve the adjustment of the blowing spacing and heating spacing of the test strip. Only when the wind speed, temperature and spacing are in the best state, the test strip will develop color the fastest and can display the extreme color in the shortest time.

[0053] Preferably, the interior of the accelerated reaction chamber 71 is hemispherical, and the inner wall of the side end of the accelerated reaction chamber 71 is provided with a thermal insulation coating, and a plurality of vertically spaced gas flow channels are evenly opened on the thermal insulation coating.

[0054] Specifically, in this embodiment, by configuring the interior of the accelerated reaction chamber 71 into a hemispherical shape, the hot air blown out of the air outlet can be rewound and circulated within the accelerated reaction chamber 71 for heating, further increasing the color development speed of the test strip. By providing an insulating coating on the inner wall of the accelerated reaction chamber 71, heat loss is reduced and energy conservation is improved. By providing vertically spaced gas flow channels, the direction of the air blown out of the air outlet can be controlled, allowing the hot air to circulate back and forth in the vertical direction.

[0055] Preferably, a timing unit 11 and a control unit 12 are provided in the test strip placement box 7, and the control unit 12 is respectively connected to the lifting mechanism, each heating resistor 76, the air inlet fan 75, each temperature detection device 77, each wind speed detection device 78 and the timing unit 11;

[0056] The timing unit 11 is used to start timing after the test strip is placed in the placement slot, and stop timing to generate the limit time after the test strip displays the limit color. The control unit 12 establishes an objective function based on the temperature, wind speed, acceleration distance and corresponding limit time at each air outlet, solves the minimum value of the objective function, and outputs the corresponding temperature, wind speed, and acceleration distance as the optimal combination when the objective function is minimum.

[0057] Specifically, in this embodiment, the timing unit 11 is used to calculate the limit time for each test strip to reach the limit color. The smaller the limit time, the better the coordination between the heating resistor 76, the air intake fan 75, and the lifting mechanism. The control unit 12 can be a computer chip. When the objective function is minimized, the temperature, wind speed, and acceleration distance at this time are recorded as the optimal combination. Based on the optimal combination, corresponding control instructions are generated and sent to the remaining lifting mechanisms, heating resistors 76, and air intake fans 75 respectively. This ensures that all heating resistors 76, air intake fans 75, and lifting mechanisms are coordinated to the best possible degree, and at this point, the color development speed of each test strip within the extreme feedback module 1 is optimized.

[0058] Preferably, the objective function is configured as:

[0059]

[0060]

[0061] Among them, f(x) is used to represent the objective function;

[0062] T i Used to indicate limit time;

[0063] T0 is used to represent the preset standard time, which is a positive number;

[0064] k1, k2, and k3 are used to represent the preset first coefficient, second coefficient, and third coefficient, respectively, and the first coefficient, second coefficient, and third coefficient are all constants;

[0065] T P Used to indicate the temperature at the air outlet;

[0066] W i Used to indicate the wind speed at the air outlet;

[0067] I n Used to represent the acceleration spacing.

[0068] Preferably, the difference function is configured as:

[0069]

[0070] Among them, D V Used to represent the difference function;

[0071] R max i is used to represent the known RGB data of the i-th pixel in the first color development area of ​​the test strip showing the extreme color;

[0072] R i Used to represent the actual RGB data of the i-th pixel in the first color development area of ​​the test strip presenting the extreme color.

[0073] Preferably, the initial model is a U-Net prediction model. In the process of retraining the U-Net prediction model based on the real-time image and the first color display area, a residual block and an adversarial network are superimposed on the U-Net prediction model. The residual block is used to improve the defect segmentation effect, and the adversarial network is used to supervise the output of the U-Net prediction model and perform feedback adjustment on the U-Net prediction model.

[0074] Specifically, this embodiment superimposes a residual block and an adversarial network on the U-Net prediction model. The residual block improves the segmentation of defect areas, reduces model bias, and improves the model's prediction accuracy. Furthermore, the adversarial network provides continuous feedback correction to the U-Net prediction model, significantly improving its accuracy.

[0075] Preferably, the content output module 6 includes:

[0076] A superposition unit 61 is used to superimpose the difference function and the content detection model to obtain a superposition function;

[0077] A calculation unit 62, configured to find the minimum value of the difference function;

[0078] The correction unit 63 is connected to the calculation unit 62 and the superposition unit 61 respectively, and is used to adjust the weight parameters of the content detection model according to the minimum value of the difference function when the minimum value of the difference function is obtained, and output the adjusted superposition function as the optimized detection model.

[0079] Specifically, in this embodiment, the correction unit 63 is used to continuously adjust the weight parameters of the content detection model, thereby avoiding the influence of color value deviation generated during the collection of test strip images, thereby improving the accuracy of human chorionic gonadotropin content detection.

[0080] Preferably, the first acquisition module 2 and the second acquisition module 4 both include preprocessing units, respectively used to eliminate invalid data, non-steady-state data and abnormal data in the test strip image to obtain preprocessed test strip data.

[0081] Specifically, in this embodiment, by providing a preprocessing unit, effective filtering of invalid data, non-steady-state data and abnormal data in the test strip image is achieved, thereby avoiding interference with subsequent operations and improving the operational efficiency of this technical solution.

[0082] The above are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions based on the principles of the present invention are within the scope of protection of the present invention. It should be noted that improvements and modifications that do not depart from the principles of the present invention are within the scope of protection of the present invention.

Claims

1. A human chorionic gonadotropin content detection system, characterized in that: include: An extreme feedback module (1) is provided with a color development acceleration device and a plurality of placement slots (72) therein, wherein the placement slots (72) are used to place human chorionic gonadotropin test strips after immunochromatography, and the color development acceleration device is used to accelerate the color development of each of the test strips until the color development area displays the limit color known by the RGB data; A first acquisition module (2) is connected to the extreme feedback module (1) and is used to acquire a test strip image of each test strip showing an extreme color and manually mark a first color-developed area; a difference calculation module (3), connected to the first acquisition module (2), for acquiring the image RGB data of each pixel in the first color display area, and establishing a difference function based on the difference between the known RGB data of the limit color and the image RGB data of each pixel; A second acquisition module (4), connected to the extreme feedback module (1), is used to acquire real-time images of each test strip and select a portion of the real-time image to manually mark the first color development area; A model training module (5) is connected to the second acquisition module (4) and is used to introduce an initial model, take the marked real-time image as input, take the marked first color area as output, and retrain the initial model to obtain a feature marking model; The content output module (6) is connected to the model training module (5) and the difference calculation module (3), and is used to input the unlabeled real-time image into the feature labeling model to obtain a feature display area, and adjust the weight parameters of the pre-trained content detection model according to the difference function to obtain an optimized detection model, and then input the RGB data of each pixel point in the feature display area into the optimized detection model to obtain the hormone content.

2. The human chorionic gonadotropin content detection system according to claim 1, characterized in that: The extreme feedback module (1) includes a test strip placement box (7), an accelerated reaction chamber (71) is provided in the test strip placement box (7), and each placement slot (72) is opened on the bottom inner wall of the accelerated reaction chamber (71). The color development acceleration device includes a disc-shaped shell (73), a ventilation shell (74) and a lifting mechanism. The upper end of the lifting mechanism is fixed on the top inner wall of the accelerated reaction chamber (71), and the driving output end of the lifting mechanism is fixedly connected to the ventilation shell (74). The lifting mechanism is used to drive the ventilation shell (74) to vertically lift. The ventilation shell (74) is provided with an air inlet channel. The air inlet An air inlet fan (75) is provided in the channel, the disc-shaped shell (73) is fixedly connected to the lower end of the ventilation shell (74), a plurality of air outlet channels are provided inside the disc-shaped shell (73), the air inlets of each of the air outlet channels are collectively connected to the air inlet channel, a heating resistance wire (76) is provided inside each of the air outlet channels, the air outlet of each of the air outlet channels faces the sample placement area of ​​each of the test strips, and the air outlet of each of the air outlet channels is provided with a temperature detection device (77), a wind speed detection device (78) and a distance detection device, which are respectively used to detect the temperature and wind speed at the air outlet and the acceleration distance between the air outlet and the test strip in real time.

3. The human chorionic gonadotropin content detection system according to claim 2, characterized in that: The interior of the accelerated reaction chamber (71) is hemispherical, and the inner wall of the side end of the accelerated reaction chamber (71) is provided with a thermal insulation coating, and a plurality of vertically spaced gas flow channels are evenly opened on the thermal insulation coating.

4. The human chorionic gonadotropin content detection system according to claim 2, characterized in that: The test strip placement box (7) is provided with a timing unit (11) and a control unit (12), and the control unit (12) is respectively connected to the lifting mechanism, each heating resistance wire (76), the air inlet fan (75), each temperature detection device (77), each wind speed detection device (78) and the timing unit (11); The timing unit (11) is used to start timing after the test strip is placed in the placement slot, and stop timing to generate a limit time after the test strip displays a limit color. The control unit (12) establishes an objective function based on the temperature, wind speed, acceleration distance and corresponding limit time at each air outlet, solves the minimum value of the objective function, and outputs the corresponding temperature, wind speed and acceleration distance as an optimal combination when the objective function is minimum.

5. The human chorionic gonadotropin content detection system according to claim 4, characterized in that: The objective function is configured as: Wherein, f(x) is used to represent the objective function; T i Used to indicate the limit time; T0 is used to represent the preset standard time, which is a positive number; k1, k2, and k3 are used to represent a preset first coefficient, a second coefficient, and a third coefficient, respectively, where the first coefficient, the second coefficient, and the third coefficient are all constants; T P Used to indicate the temperature at the air outlet; W i Used to indicate the wind speed at the air outlet; I n Used to represent the acceleration spacing.

6. The human chorionic gonadotropin content detection system according to claim 1, characterized in that: The difference function is configured as follows: Among them, D V Used to represent the difference function; R maxi Known RGB data for representing the i-th pixel of the test strip in the first color development area that presents the extreme color; R i Used to represent actual RGB data of the i-th pixel in the first color development area of ​​the test strip presenting the extreme color.

7. The human chorionic gonadotropin content detection system according to claim 1, characterized in that: The initial model is a U-Net prediction model. In the process of retraining the U-Net prediction model based on the real-time image and the first color display area, a residual block and an adversarial network are superimposed on the U-Net prediction model. The residual block is used to improve the defect segmentation effect, and the adversarial network is used to supervise the output of the U-Net prediction model and perform feedback adjustment on the U-Net prediction model.

8. The human chorionic gonadotropin content detection system according to claim 1, characterized in that: The content output module (6) comprises: A superposition unit (61) is used to superimpose the difference function and the content detection model to obtain a superposition function; A calculation unit (62) for solving the minimum value of the difference function; The correction unit (63) is connected to the calculation unit (62) and the superposition unit (61) respectively, and is used to adjust the weight parameters of the content detection model according to the minimum value of the difference function when the minimum value of the difference function is obtained, and output the adjusted superposition function as the optimized detection model.

9. The human chorionic gonadotropin content detection system according to claim 1, characterized in that: The first acquisition module (2) and the second acquisition module (4) both include preprocessing units, which are respectively used to eliminate invalid data, non-steady-state data and abnormal data in the test strip image to obtain the preprocessed test strip data.

Images